Streammetabolizer output for GBU
Kelly Loria
2025-02-04
Background
This is the upper reach of a small, turbulent, shaded, stream with a fair amount of fine particulate organic matter.
Identifying reasonable modeled estimates of K600
We then ran the normal stream metabolizer model:
b_Kb_oipi_tr_plrckm.stan to get modeled K600 to see if we
could resolve the negative correlation between ER and K600. Priors on
K600_lnQ_nodes_meanlog were set as 5 bins based on
rnorm(1000, mean = logQ_mean, sd = logQ_sd) centered around
the mean and logQ va;ues 1-2 sd away from the mean.
Load different model segments with normal prior for gas exchange.
We chose segments of time where we believe GPP occurred and was greater than 0. These chunks of time are from a previous model where we binned flow and incorporated measured and estimated K600 priors from gas exchange measurements a the reach.
Metabolism with CIs for the full time series
This the raw model output. It looks okay aside from that part of 2023 where ER strangely dips. Where GPP is in blue and ER is in orange.
Here is the run configuration for full model:
mm_name(type = 'bayes', pool_K600 = "binned", err_obs_iid = TRUE, err_proc_iid = TRUE, ode_method = "trapezoid", deficit_src = 'DO_mod', engine = 'stan')
Fitting priors:
K600_lnQ_nodes_meanlog = log(32) Where 32 was the mean
value from observed measurements and normal pooled modeled,
K600_lnQ_nodes_sdlog = 1.2 was 1.2,
bayes_specs_new$K600_lnQ_nodes_centers <- log_bins was
from
prior_samples <- rnorm(1000, mean = logQ_mean, sd = logQ_sd)
Checking full model convergence
Make sure the chains converged; all r-hat values were well below 1.05 (the red line) for GPP, ER, and K600. The blue lines are the mean for each parameter.
** Not all all segments had great convergence.
Looks like that weird 2023 time period corresponds to bad rhats for all parameters.
## [1] 1.011847## [1] 1.001453## [1] 1.008392
## date lab rmse sd
## Min. :2021-05-01 Length:120 Min. :0.02600 Min. :0.08183
## 1st Qu.:2021-05-08 Class :character 1st Qu.:0.05183 1st Qu.:0.36078
## Median :2021-05-23 Mode :character Median :0.09201 Median :0.43503
## Mean :2021-08-15 Mean :0.11585 Mean :0.42122
## 3rd Qu.:2021-09-03 3rd Qu.:0.13302 3rd Qu.:0.46960
## Max. :2022-05-15 Max. :0.43387 Max. :0.75359
## NA's :16
## min max range nrmse
## Min. :6.022 Min. : 7.802 Min. :0.2227 Min. :0.02340
## 1st Qu.:7.430 1st Qu.: 8.801 1st Qu.:1.2165 1st Qu.:0.03977
## Median :7.738 Median : 9.033 Median :1.3600 Median :0.06643
## Mean :7.773 Mean : 9.143 Mean :1.3699 Mean :0.07587
## 3rd Qu.:8.190 3rd Qu.: 9.479 3rd Qu.:1.5277 3rd Qu.:0.10308
## Max. :9.612 Max. :10.815 Max. :2.8733 Max. :0.18608
## NA's :16
## minT maxT rangeT
## Min. : 1.790 Min. : 5.53 Min. :0.525
## 1st Qu.: 5.617 1st Qu.:11.55 1st Qu.:5.524
## Median : 6.961 Median :12.90 Median :6.033
## Mean : 6.966 Mean :12.72 Mean :5.750
## 3rd Qu.: 8.165 3rd Qu.:14.25 3rd Qu.:6.559
## Max. :11.603 Max. :17.25 Max. :8.051
## Looking at the estimated K600 and ER relationships from normal pooling
Here is the run configuration to get modeled K600:
mm_name(type = 'bayes', pool_K600 = "normal", err_obs_iid = TRUE, err_proc_iid = TRUE, ode_method = "trapezoid", deficit_src = 'DO_mod', engine = 'stan')
Where each “lab” segment was run as individual model with the
K600_lnQ_nodes_meanlog adjusted to match streamline in cms
during that time.
K-ER, K-Q, general K600 trends from segements with GPP > 0
There is a strong negative correlation between ER and K600 (-0.828). GPP and K600 are slightly less correlated (0.013). And Likely a positive relationship between flow and K600.
## [1] 32.85907## [1] 39.53105## [1] 36.19506KER_cor <- round(cor(met.df$ER_daily_mean, met.df$K600_daily_mean, use = "complete.obs"),3)
print(KER_cor)## [1] -0.828KGPP_cor <-round(cor(met.df$GPP_daily_mean, met.df$K600_daily_mean, use = "complete.obs"),3)
print(KGPP_cor)## [1] 0.013
The vertical dashed is the overall mean modeled K600 in the box plot.
Compared modeled and measured K600
Plots for (1) measured v modeled K600 and flow and (2) logK600 and log(flow+1).
Could be one poor measurement at the highest flow for measured gas
exchange. But in general the modeled K600 does seem similar to the
measured, which is kind of nice to see how robust the
pool_K600 = "normal" is getting at K600.
Full run with binned K600 informed by both measured and modeled priors
Here is the run configuration for full model:
mm_name(type = 'bayes', pool_K600 = "binned", err_obs_iid = TRUE, err_proc_iid = TRUE, ode_method = "trapezoid", deficit_src = 'DO_mod', engine = 'stan')
dat1 <- input_dat%>% filter(discharge<0.99)
## Set bayes specs
bayes_name_new <- mm_name(type = 'bayes', pool_K600 = "binned",
err_obs_iid = TRUE, err_proc_iid = TRUE,
ode_method = "trapezoid", deficit_src = 'DO_mod', engine = 'stan')
bayes_specs_new <- specs(bayes_name_new)
# Compute log-transformed discharge
logQ <- log(na.omit(dat1$discharge))
sd_logQ <- round(sd(na.omit(logQ)),2)
## Compute log-transformed K600 for prior estimation
logQ_mean <- round(mean(logQ, na.rm = TRUE),2) # Center the prior flow mean
logQ_mean## [1] -3.94## [1] 1.5## [1] 3.5## [1] 0.93## [1] 83.93142## [1] 13.06582prior_samples <- rnorm(1000, mean = logQ_mean, sd = logQ_sd)
# Define bins based on log-scale mean and SD
log_bins <- c(logQ_mean - c(2 * logQ_sd),
logQ_mean - c(1 * logQ_sd),
logQ_mean,
logQ_mean + c(1 * logQ_sd),
logQ_mean + c(2 * logQ_sd))
# Plot:
#bins <- exp(log_bins)
prior_df <- data.frame(Q_cms = prior_samples)
prior_bins_plot <- ggplot(prior_df, aes(x = Q_cms)) +
geom_density(fill = "lightblue", alpha = 0.5) + # Density plot
geom_vline(xintercept = log_bins, color = "black", lty="dashed") + # Bin edges
labs(title = "Binned log(flow) for K600 Priors", y = "Density", x = "log(Q) (cms)") +
theme_bw()
prior_bins_plot
# Assign priors for binned K600
bayes_specs_new$K600_lnQ_nodes_meanlog <- rep(logK600_mean, length(log_bins)) # Centered at 25
bayes_specs_new$K600_lnQ_nodes_meanlog## [1] 3.5 3.5 3.5 3.5 3.5bayes_specs_new$K600_lnQ_nodes_sdlog <- rep(logK600_sd, length(log_bins)) # Wide enough to allow 40
bayes_specs_new$K600_lnQ_nodes_sdlog ## [1] 0.93 0.93 0.93 0.93 0.93## [1] -6.94 -5.44 -3.94 -2.44 -0.94# Keep other parameters
bayes_specs_new$K600_daily_sigma_sigma <- 0.05
bayes_specs_new$n_chains <- c(3)
bayes_specs_new$n_cores <- c(3)
bayes_specs_new$burnin_steps <- c(2500)
bayes_specs_new$saved_steps <- c(2500)
bayes_specs_new## Model specifications:
## model_name b_Kb_oipi_tr_plrckm.stan
## engine stan
## split_dates FALSE
## keep_mcmcs TRUE
## keep_mcmc_data TRUE
## day_start 4
## day_end 28
## day_tests full_day, even_timesteps, complete_data, pos_disch...
## required_timestep NA
## K600_lnQ_nodes_centers -6.94, -5.44, -3.94, -2.44, -0.94
## GPP_daily_mu 3.1
## GPP_daily_lower -Inf
## GPP_daily_sigma 6
## ER_daily_mu -7.1
## ER_daily_upper Inf
## ER_daily_sigma 7.1
## K600_lnQ_nodediffs_sdlog 0.5
## K600_lnQ_nodes_meanlog 3.5, 3.5, 3.5, 3.5, 3.5
## K600_lnQ_nodes_sdlog 0.93, 0.93, 0.93, 0.93, 0.93
## K600_daily_sigma_sigma 0.05
## err_obs_iid_sigma_scale 0.03
## err_proc_iid_sigma_scale 5
## params_in GPP_daily_mu, GPP_daily_lower, GPP_daily_sigma, ER...
## params_out GPP, ER, DO_R2, GPP_daily, ER_daily, K600_daily, K...
## n_chains 3
## n_cores 3
## burnin_steps 2500
## saved_steps 2500
## thin_steps 1
## verbose FALSEPlots made on filtered data: met.clean
filtered for days with
GPP_daily_Rhat<1.05,ER_daily_Rhat<1.05,
K600_daily_Rhat <1.05, as well as
(GPP_97.5pct>0) and (ER_2.5pct<0).
met.clean <- met.full %>%
filter(GPP_daily_Rhat<1.05)%>%
filter(GPP_97.5pct>0)%>%
filter(ER_daily_Rhat<1.05) %>%
filter(ER_2.5pct<0)%>%
filter(K600_daily_Rhat<1.05) %>%
filter(K600_daily_mean<45)
mean_k_mod <- mean(met.clean$K600_daily_mean)
mean_k_mod## [1] 20.02085## [1] 39.53105KER_cor <- round(cor(met.clean$ER_daily_mean, met.clean$K600_daily_mean, use = "complete.obs"),3)
print(KER_cor)## [1] -0.763KGPP_cor <-round(cor(met.clean$GPP_daily_mean, met.clean$K600_daily_mean, use = "complete.obs"),3)
print(KGPP_cor)## [1] -0.062
GPP and ER
## [1] -0.125
Final thoughts.
The direction of the K600 ~ flow relationship looks more logical, where K600 increases with flow. ER and K600 are still negatively correlated (-0.732) but, less strongly relative to the lower reach (GBL). The relationship between K600 and flow appears to be positive but still inflected in a strange way.
However I’m still think we should be cautious in over interpreting ER trends.
The mean modeled K600 is lower than expected, 20 when it should be closer to 32. There was also a chunk of time with poor rhats in 2023 for all parameters where K600 was greater than 45, that was removed for this analysis.
Final cleaned model output:
Where GPP is in blue and ER is in orange, and the black points represent NEP.
Table summarizing excluded data from cleaned model output:
Of the 933 days with DO observations 474 days were removed.
| Number of Days | Explaination | Percent of Days |
|---|---|---|
| 933 | Total days of DO observations | 100.0 |
| 474 | Total days removed | 50.8 |
| 126 | Days model was unable to fit | 13.5 |
| 0 | days where GPP rhat > 1.05 | 0.0 |
| 0 | days where ER rhat > 1.05 | 0.0 |
| 0 | days where K600 rhat > 1.05 | 0.0 |
| 101 | days where modeled GPP was negative | 10.8 |
| 5 | days where modeled ER was positive | 0.5 |
Session info
R version 4.4.2 (2024-10-31)
Platform: aarch64-apple-darwin20
locale: en_US.UTF-8||en_US.UTF-8||en_US.UTF-8||C||en_US.UTF-8||en_US.UTF-8
attached base packages: stats, graphics, grDevices, utils, datasets, methods and base
other attached packages: plotly(v.4.10.4), kableExtra(v.1.4.0), knitr(v.1.49), streamMetabolizer(v.0.12.1), ggpubr(v.0.6.0), readxl(v.1.4.3), zoo(v.1.8-12), cowplot(v.1.1.3), viridis(v.0.6.5), viridisLite(v.0.4.2), dataRetrieval(v.2.7.17), lubridate(v.1.9.4), forcats(v.1.0.0), stringr(v.1.5.1), dplyr(v.1.1.4), purrr(v.1.0.2), readr(v.2.1.5), tidyr(v.1.3.1), tibble(v.3.2.1), ggplot2(v.3.5.1) and tidyverse(v.2.0.0)
loaded via a namespace (and not attached): DBI(v.1.2.3), gridExtra(v.2.3), rlang(v.1.1.4), magrittr(v.2.0.3), e1071(v.1.7-16), compiler(v.4.4.2), mgcv(v.1.9-1), systemfonts(v.1.1.0), vctrs(v.0.6.5), pkgconfig(v.2.0.3), crayon(v.1.5.3), fastmap(v.1.2.0), backports(v.1.5.0), labeling(v.0.4.3), pander(v.0.6.5), deSolve(v.1.40), rmarkdown(v.2.29), tzdb(v.0.4.0), bit(v.4.5.0.1), xfun(v.0.49), cachem(v.1.1.0), jsonlite(v.1.8.9), broom(v.1.0.7), parallel(v.4.4.2), R6(v.2.5.1), bslib(v.0.8.0), stringi(v.1.8.4), car(v.3.1-3), jquerylib(v.0.1.4), cellranger(v.1.1.0), Rcpp(v.1.0.14), Matrix(v.1.7-1), splines(v.4.4.2), timechange(v.0.3.0), tidyselect(v.1.2.1), rstudioapi(v.0.17.1), abind(v.1.4-8), yaml(v.2.3.10), lattice(v.0.22-6), plyr(v.1.8.9), withr(v.3.0.2), evaluate(v.1.0.1), rLakeAnalyzer(v.1.11.4.1), sf(v.1.0-19), units(v.0.8-5), proxy(v.0.4-27), xml2(v.1.3.6), pillar(v.1.10.1), carData(v.3.0-5), KernSmooth(v.2.23-24), generics(v.0.1.3), vroom(v.1.6.5), hms(v.1.1.3), munsell(v.0.5.1), scales(v.1.3.0), class(v.7.3-22), glue(v.1.8.0), lazyeval(v.0.2.2), tools(v.4.4.2), data.table(v.1.16.4), ggsignif(v.0.6.4), LakeMetabolizer(v.1.5.5), grid(v.4.4.2), crosstalk(v.1.2.1), colorspace(v.2.1-1), nlme(v.3.1-166), Formula(v.1.2-5), cli(v.3.6.3), svglite(v.2.1.3), gtable(v.0.3.6), rstatix(v.0.7.2), sass(v.0.4.9), digest(v.0.6.37), classInt(v.0.4-11), htmlwidgets(v.1.6.4), farver(v.2.1.2), htmltools(v.0.5.8.1), lifecycle(v.1.0.4), httr(v.1.4.7), unitted(v.0.2.9) and bit64(v.4.5.2)